mcp-scraper 0.61.0 → 0.62.0

package/README.md

@@ -90,7 +90,7 @@ Build the branded one-click bundle:
 npm run build:mcpb
 ```
 
-The generated bundle is written to `build/mcpb/mcp-scraper-<version>.mcpb` and copied to `public/downloads/` for the hosted download. The current public bundle is `https://mcpscraper.dev/downloads/mcp-scraper.mcpb` (`0.61.0`, SHA-256 `69d0a740c2b75b449deb910d0fc2085dd7b68dc4e7f10bbf0de8b22d4fb221d3`). Install it by opening or dragging it into Claude Desktop. Claude displays the `MCP Scraper` install card, icon, and API-key configuration field from the bundle manifest.
+The generated bundle is written to `build/mcpb/mcp-scraper-<version>.mcpb` and copied to `public/downloads/` for the hosted download. The current public bundle is `https://mcpscraper.dev/downloads/mcp-scraper.mcpb` (`0.62.0`, SHA-256 `83a756b2a3c8acd07813a8ac475b20b02bf56004fb6c207292da67ac3d76a9f4`). Install it by opening or dragging it into Claude Desktop. Claude displays the `MCP Scraper` install card, icon, and API-key configuration field from the bundle manifest.
 
 The MCPB install exposes every tool — web-intelligence plus all `browser_*` tools — through the one `mcp-scraper` server.
 

package/dist/bin/api-server.cjs

@@ -44889,7 +44889,7 @@ var PACKAGE_VERSION;
 var init_version = __esm({
   "src/version.ts"() {
     "use strict";
-    PACKAGE_VERSION = "0.61.0";
+    PACKAGE_VERSION = "0.62.0";
   }
 });
 
@@ -50456,7 +50456,7 @@ function registerPaaExtractorMcpTools(server, executor, options = {}) {
   }, async (input) => formatSearchSerp(await executor.searchSerp(input), input));
   server.registerTool("extract_url", {
     title: "Single URL Extract",
-    description: "Extract structured data from one public URL: content, schema, headings, metadata, screenshots, branding, featured image, or media assets. Wayback replay URLs automatically return the archived page copy without playback chrome. Use delivery:auto for bounded inline results with automatic artifact offload, delivery:artifact for a durable owner-scoped report, or delivery:memory to save the full page into hosted MCP Memory. preserveMedia is the preferred media-retention flag; depositToVault and downloadMedia remain temporary compatibility aliases.",
+    description: "Extract structured data from one public URL: content, schema, headings, metadata, screenshots, branding, featured image, or media assets. Tries a plain HTTP fetch first and only opens a stealth browser when that fetch is blocked, hits a bot check, or returns unusably thin content \u2014 most calls never need the browser, and a bot_check_unresolved error means the browser already tried and failed, not that a retry will help. Wayback replay URLs automatically return the archived page copy without playback chrome. Use delivery:auto for bounded inline results with automatic artifact offload, delivery:artifact for a durable owner-scoped report, or delivery:memory to save the full page into hosted MCP Memory. preserveMedia is the preferred media-retention flag; depositToVault and downloadMedia remain temporary compatibility aliases.",
     inputSchema: exposesLocalNetworkAccess ? ExtractUrlLocalInputSchema : ExtractUrlInputSchema,
     outputSchema: recordOutputSchema("extract_url", ExtractUrlOutputSchema),
     annotations: { ...liveWebToolAnnotations("Single URL Extract"), readOnlyHint: false }
@@ -50484,7 +50484,7 @@ function registerPaaExtractorMcpTools(server, executor, options = {}) {
   }, async (input) => formatMapWaybackSnapshots(await executor.mapWaybackSnapshots(input), input, ctx));
   server.registerTool("extract_site", {
     title: "Multi-Page Site Content Crawl",
-    description: `Crawl a public website and return page CONTENT (Markdown) across multiple pages. A Wayback replay URL produces one archived site snapshot. The optional wayback plan produces whole-site, single-page, or selected-page timelines across explicit months or a month range, all in one export with a capture matrix. Pass a new idempotencyKey for each intended crawl and reuse it only when retrying that call. Every MCP crawl starts a durable export; poll check_site_export for honest outcome counters and ${fileBehavior("the saved ZIP.", "the owner-scoped downloadable ZIP.")} Content only \u2014 for a technical SEO audit use audit_site instead.`,
+    description: `Crawl a public website and return page CONTENT (Markdown) across multiple pages. Each page is fetched with a plain HTTP request first; a browser retries only the pages that were blocked or hit a bot check, so a per-page bot_check_unresolved result means that specific page already failed the browser retry too. A Wayback replay URL produces one archived site snapshot. The optional wayback plan produces whole-site, single-page, or selected-page timelines across explicit months or a month range, all in one export with a capture matrix. Pass a new idempotencyKey for each intended crawl and reuse it only when retrying that call. Every MCP crawl starts a durable export; poll check_site_export for honest outcome counters and ${fileBehavior("the saved ZIP.", "the owner-scoped downloadable ZIP.")} Content only \u2014 for a technical SEO audit use audit_site instead.`,
     inputSchema: ExtractSiteInputSchema,
     outputSchema: recordOutputSchema("extract_site", ExtractSiteOutputSchema),
     annotations: { ...liveWebToolAnnotations("Multi-Page Site Content Crawl"), readOnlyHint: false }